About this episode
Anthropic drops Opus 4.6. Twenty minutes later, OpenAI fires back with GPT-5.3 Codex. This is the AI agentic coding arms race and it's moving fast. Both AI models are writing code that can write itself now. OpenAI is using 5.3 to improve its own tooling. Opus 4.6 is "voicing discomfort with being a product." We tested both and break down what actually matters for people building stuff. Plus Kling 3.0 is out (and harder to prompt than you think), OpenClaw bots are hiring humans on rent-a-human.ai, Roblox launches prompt-to-3D creation, and robots are now doing 130K step challenges in negative 47 degree weather. THE MODELS ARE IMPROVING THEMSELVES NOW. EVERYTHING IS FINE. Come to our Discord: https://discord.gg/muD2TYgC8f Join our Patreon: https://www.patreon.com/AIForHumansShow AI For Humans Newsletter: https://aiforhumans.beehiiv.com/ Follow us for more on X @AIForHumansShow Join our TikTok @aiforhumansshow To book us for speaking, please visit our website: https://www.aiforhumans.show/ // Show Links // Anthropic's Claude Opus 4.6 https://www.anthropic.com/news/claude-opus-4-6 Orchestrating Agents in Claude Code https://x.com/lydiahallie/status/2019469032844587505?s=20 Opus 4.6 Beats Humans at analyzing complex human science docs https://x.com/_simonsmith/status/2019502742209769540?s=20 OpenAI GPT-5.3 Codex https://openai.com/index/introducing-gpt-5-3-codex/ 5.3 Codex First model instrumental in creating itself https://x.com/deredleritt3r/status/2019475360438493597 OpenAI Frontier https://openai.com/index/introducing-openai-frontier/ Anthropic's Superbowl Ads https://x.com/tomwarren/status/2019039874771550516?s=20 GPT-5 connected to an autonomous lab to do experiments https://x.com/OpenAI/status/2019488071134347605?s=20 OpenClaw https://openclaw.ai/ Rent-A-Human https://rentahuman.ai/bounties Kling 3.0 = Really good model https://x.com/Kling_ai/status/2019064918960668819?s=20 Kling 3.0 Moonlanding Mockumentary
